The "counter"

[Battlemuffins 26]

There seems to be an issue among the testers with the fact that 'page views' are counted instead of 'play counts'. I personally have an issue with this becuase it would take no effort for me to gather 5-10 friends and have them spam F5 until I reach the front page. But aside from that, a page view doesn't truly count as a play. One could simply visit the page to click a link or read the description, or even simply to admire album art. While this is not a HUGE issue, I still feel it one that could be looked into.

[Anorax 29]

Diagram tiem!

 

sweet bro and hella jeff style

 

 

why play cont is usful

 

ok, enough with the comic sans

 

while page spamming may not be an issue, it may not accurately display the "play" count a song gets

 

This is especially true now that Pony.FM has external embeds in the MLPforums

 

another way that the "views" may not equal "plays" is where the page is loaded, not to play the song, but to read the description or download the song if someone has already listened to the song.

 

While view-only results work well with sites like YouTube, YouTube counts "views" as "plays" - when the video starts loading and playing, it counts a view - not when the page loads (this is especially true if YouTube has a way for the video uploader to disable "autoplay on page load")

 

just my 2 cents on this

Edited January 4, 2013 by Anorax

[CloudFyre 848]

I've mentioned this to Feld0 before, and he had some varied thoughts on the subject. For one, I believe he mentioned that most people wouldn't bother with spamming F5. Don't quote me on that, but I think there will be some rather unscrupulous users who will eventually attempt it.

I'm with you guys:

"Page visit" = 1 page visit.
"Play" = 1 click of the "play" button

"Download" = 1 unique download of a song in a single format.

[Twilight Sparkle ✨ 8,520]

I've got a play counter on my list to add today. We run into a minor semantics problem, though: should a "play" be logged only once per page view, or should it be logged again if you listen through the track and then restart it? If a play is logged every time the play button is clicked, one could spam up the counter by simply going clickfire on the play/pause button.

 

 

Regarding F5 spamming... that's completely possible on deviantART. It hasn't been an issue there, has it?

[CloudFyre 848]

I've got a play counter on my list to add today. We run into a minor semantics problem, though: should a "play" be logged only once per page view, or should it be logged again if you listen through the track and then restart it? If a play is logged every time the play button is clicked, one could spam up the counter by simply going clickfire on the play/pause button.

 

 

Regarding F5 spamming... that's completely possible on deviantART. It hasn't been an issue there, has it?

 

I'd say 1 play would be the max per 1 page view, as I could easily see someone needing to stop a song momentarily (phone call, parents, etc.) and then coming back to resume listening.

 

On the other hand though...I listen to some songs over and over because I like them. I'm not really sure what to think.

 

If you could implement some kind of tracker that kept tabs on how long a song has been running, then you would have the power to do a more complex counter system.

 

 

For an easy method, 1 click of the play button = 1 "play", and the song has to end before another play can be counted.

 

If you wanted something more complex:

1 click of the play button = 1 "play", so long as a listener is tuning in for at least 15 (30?) seconds total. (The time formula would have to be altered to a ratio, in case someone uploads short intros/outros/sfx/etc.) After the minimum time is reached, a play count is added, and the song must reach it's end in order for the playcount code to reset. (This is the approach that YouTube takes to determine whether a video is getting real "views" or not.)

 

I actually wrote up a basic (undetailed) Python code for the above process, but I decided not to include it in favor of the word description above.

Edited January 4, 2013 by CloudFyre

[Twilight Sparkle ✨ 8,520]

I just wrapped up a basic play counter. It will add one play to the count per completion of the song.

 

Currently, doing this also adds one download to the download count, because the site's player grabs the standard MP3 download. I'll add a secondary URL to access the MP3 that doesn't log a download to correct this.

Update: A "play" no longer adds a "download".

Update 2: I changed the "popular tracks" algorithm to retrieve tracks with the most plays in the past 24 hours. Prior, the metric used was views. You can see the results on the homepage and the Today's Top 20 page.

[Bohtty 1,775]

What would be the viability of limiting views by IP address? I think that views and plays will end up being very similar for most people since usually someone would click on a song to listen to the song. If both counters read similar or the same then I think that it would look worse due to the needless repetition. I feel it would be more efficient to limit view counter abuse or replace view counter with play counter.

 

Additionally if you click the random track button and it brings you to a song you don't want then that view is meaningless which would decrease the usefulness of the view counter I would think.

[Twilight Sparkle ✨ 8,520]

What would be the viability of limiting views by IP address? I think that views and plays will end up being very similar for most people since usually someone would click on a song to listen to the song. If both counters read similar or the same then I think that it would look worse due to the needless repetition. I feel it would be more efficient to limit view counter abuse or replace view counter with play counter.

 

Views, plays, and downloads all log your IP address, so that can definitely be used for something. I'm not so sure that it should, though; you can get a play without getting a corresponding view from an embedded player (on a non-Pony.fm site) or an inline player (like the tracks listing or an album page).

 

 

 

Additionally if you click the random track button and it brings you to a song you don't want then that view is meaningless which would decrease the usefulness of the view counter I would think.

 

Not necessarily. If a track has more views than plays, would it not be an indicator of a lack of interest in it? And if it has more plays than views, an indicator that people enjoy listening to it many times?